Chimpanzees Deceive a Human Competitor_06

7/30/2019 Chimpanzees Deceive a Human Competitor_06

1/20

Chimpanzees deceive a human competitorby hiding

Brian Hare a,b,*, Josep Call b, Michael Tomasello b

a

Department of Anthropology, Harvard University, Cambridge, MA 02138, USAb Max-Planck-Institute for Evolutionary Anthropology, Inselstrasse 22, D-04103 Leipzig, Germany

Received 19 January 2004; revised 30 July 2004; accepted 14 January 2005

Abstract

There is little experimental evidence that any non-human species is capable of purposefully

attempting to manipulate the psychological states of others deceptively (e.g., manipulating

what another sees). We show here that chimpanzees, one of humans two closest primate rel-

atives, sometimes attempt to actively conceal things from others. Specifically, when competing

with a human in three novel tests, eight chimpanzees, from their first trials, chose to approacha contested food item via a route hidden from the humans view (sometimes using a circuitous

path to do so). These findings not only corroborate previous work showing that chimpanzees

know what others can and cannot see, but also suggest that when competing for food chim-

panzees are skillful at manipulating, to their own advantage, whether others can or cannot see

them.

2006 Published by Elsevier B.V.

Keywords: Chimpanzee; Deception; Visual perspective taking; Social cognition; Cognitive evolution

1. Introduction

Human beings sometimes attempt to deceive one another. Whereas various relat-

ed but non-mentalistic phenomena such as bodily camouflage are widespread in the

0010-0277/$ - see front matter 2006 Published by Elsevier B.V.

doi:10.1016/j.cognition.2005.01.011

* Corresponding author.

E-mail address: [email protected] (B. Hare).

www.elsevier.com/locate/COGNITCognition 101 (2006) 495514
mailto:[email protected]:[email protected]


2/20

animal kingdom, intentional deception in which one individual attempts to actively

manipulate what another experiences cognitively is considered by many to be a

uniquely human cognitive ability (e.g., Cheney & Seyfarth, 1990; Hauser, 1997;Tomasello & Call, 1997).

While research on non-human primates has revealed a number of behaviors that

function to deceive others (so-called tactical deception, Whiten & Byrne, 1988), it

remains unclear whether such cases also represent instances of intentional deception.

For instance, subordinate primates refrain from giving food calls that might attract

dominants to monopolizable food (Hauser, 1992, 1997; Hauser, Teixidor, Field, &

Flaherty, 1993), learn to hide themselves from potential competitors (de Waal,

1998; Gygax, 2000; Kummer, Anzenberger, & Hemelrijk, 1996), and sometimes even

actively lead approaching dominants away from the location of hidden food (Coussi-

Korbel, 1994; Goodall, 1986; Hirata & Matsuzawa, 2001; Menzel, 1974). However,

in each of these cases it is plausible that the animals are relying inflexibly on some

invariant behavioral or contextual cue (Cheney & Seyfarth, 1990; Heyes, 1998;

Ristau, 1991). For example, it is possible that subordinate primates simply learned

during daily competition over food that they were more likely to obtain food if they

refrained from behaving towards it in the presence of a dominant (Cheney &

Seyfarth, 1990; Heyes, 1998).

Support for such a learning interpretation comes from experiments in which it

took many dozens or even hundreds of trials for primates to learn to refrain from

indicating the location of hidden food when an individual who did not share food

was present (Anderson, Kuroshima, Kuwahata, Fujita, & Vick, 2001; Fujita, Kuro-

shima, & Masuda, 2002; Mitchell & Anderson, 1997; Woodruff & Premack, 1979).

Most notably, Woodruff and Premack (1979) trained four chimpanzees to indicate

the location of hidden food for humans, and then introduced two different experi-menters. One experimenter shared the food with the subject if he found it while

the second experimenter (wearing a mask) did not share the food with the subject

if he found it. Only two of the chimpanzees eventually learned, after dozens of trials,

to refrain from indicating the correct food location in the presence of the masked

experimenter. Because it takes primates so long to master such skills and their newly

acquired skills do not easily generalize to new settings, it is possible that in these

studies primates base their decisions on invariant contextual or behavioral cues

learned in each experimental setting (i.e., the chimpanzees of Woodruff & Premack,

1979 learned to use the mask as a discriminative cue to withhold signaling). It is

based on findings such as these, that the majority of theorist have concluded that

there is no solid experimental evidence demonstrating that non-human primates,or other animals, are capable of intentional deception (Hauser, 1997; Heyes, 1998;

Povinelli, 2000; Ristau, 1991; Shettleworth, 1998; Tomasello & Call, 1997).

However, recent studies with one of humans two closest primate relatives, the

chimpanzee, suggest the possibility for a new test of the hypothesis that intentional

deception is unique to humans. A powerful test might be possible since a recent series

of experiments demonstrate that chimpanzees understand what other individuals can

and cannot see and in certain situations what they have and have not seen in

the immediate past (Call, Agnetta, & Tomasello, 2000; Hare, Call, Agnetta, &

496 B. Hare et al. / Cognition 101 (2006) 495514
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


3/20

Tomasello, 2000; Hare, Call, & Tomasello, 2001; Tomasello, Hare, & Agnetta, 1999;

Tomasello, Hare, & Fogleman, 2001). For example, subordinate chimpanzees prefer

to retrieve food that is hidden behind an occluder and out of a dominant competi-tors view rather than food that is visible to both even if in each contest the sub-

ordinate must make her decision before seeing the dominant or the dominant

behaving (Hare et al., 2000). In addition, subordinate chimpanzees are more likely

to approach food if the dominant did not see than if the dominant did see the food

hidden behind one of two opaque occluders (Hare et al., 2001). These findings sug-

gest, in the right context (i.e., during natural food competition with conspecifics that

does not involve using arbitrary cues such as masks to indicate deceitfulness), chim-

panzees, like humans, might also demonstrate skill at actively and flexibly manipu-

lating what others can and cannot see when it is to their benefit.

But there remains debate regarding whether these chimpanzee social problem

solving behaviors might also be explained by an inflexible reliance on contextual or

behavioral cues alone (Povinelli & Vonk, 2003; Tomasello, Call, & Hare, 2003a).

Karin-Karin-DArcy and Povinelli (2002) suggest that the behavior of subordinates

in Hare et al. (2000) represents inflexible preferences for approaching food near the

safety of physical barriers during competition (although see Tomasello et al.,

2003a, Tomasello, Call, & Hare, 2003b) and not an understanding of their occluding

properties. Meanwhile, the previous findings of Povinelli and colleagues suggest that

chimpanzees are extremely limited in their understanding of how others visual

perception actually works (i.e., they show little skill in using social cues related to

others visual perception such as face direction in determining when someone can or

cannot see them) (Povinelli, Bierschwale, & Cech, 1999; Povinelli & Eddy, 1996;

Reaux, Theall, & Povinelli, 1999; Theall & Povinelli, 1999; however, see Leavens &

Hopkins, 1998; Hostetter, Contera, & Hopkins, 2001; Kaminski, Call, & Tomasello,2004; Myowa-Yamakoshi, Tomonaga, Tanaka, & Matsuzawa, 2003 for contradicto-

ry evidence).

Therefore, in the current investigation we examine whether chimpanzees are not

only capable of assessing when a competitor can and cannot see things, but also

whether they use this same ability to intentionally manipulate another individuals

visual information by concealing their approach toward contested food. In doing

so, this investigation was designed to test the ability of the visual perspective-taking

hypothesis ofHare et al. (2000, 2001) against that of the learned cues hypotheses of

Povinelli and Eddy (1996) and Karin-Karin-DArcy and Povinelli (2002) in predict-

ing the behavior of chimpanzees in solving several novel social problems.

Because chimpanzees perform most skillfully in competitive cognitive tasks (Hare,2001; Hare & Tomasello, 2004), subjects competed against a human (E) who moved

prized food items out of the subjects reach when they attempted to retrieve them.

Subjects spontaneous behavior was measured in three novel experimental setups.

In the first, subjects chose to either approach food that Es face was oriented toward

or food that Es chest (but not face) was oriented toward. In the second, subjects

chose to approach food either from behind a transparent barrier or from behind

an opaque occluder as E faced the subject. In the third, subjects chose to approach

from behind a barrier that occluded their approach either partially or fully as E faced

B. Hare et al. / Cognition 101 (2006) 495514 497


4/20

the subject. Unlike previous studies (i.e., Woodruff & Premack, 1979), this compet-

itive paradigm did not require any behavioral shaping (i.e., training a new behavior)

or communication in which the subject had to produce (or not) communicative sig-nals arguably a more difficult domain as it involves not just reading intentions but

reading communicative intentions (Tomasello, 1999; Hare & Tomasello, 2004).

Thus, in the current experiment, the visual perspective-taking hypothesis predicts

that subjects will spontaneously conceal their approach from their competitor by (1)

avoiding food that Es face is oriented toward over food that Es chest is oriented

toward (first experimental setup), (2) preferring to approach behind visual occluders

over non-occluding barriers (second and third experimental setups), and (3) avoiding

a direct approach to the food if a more indirect route might better conceal their

intent to approach (first and second setups). Meanwhile, the learned cues hypothesis

predicts that subjects inflexible use and limited understanding of social cues related

to visual perception (Povinelli, 2000; Povinelli & Eddy, 1996) and attraction to

physical barriers (regardless of their occluding properties) during food competition

(Karin-Karin-DArcy & Povinelli, 2002; Povinelli & Giambrone, 2001) will prohibit

them from concealing their approach in any of these ways.

2. Method

2.1. Subjects

Eight chimpanzees participated and could stop participating at any time (see

Table 1 for the sex, age, and rearing history of each subject). The chimpanzees live

at the Wolfgang Kohler Primate Research Center (4533 m2) in the Leipzig Zoo, Ger-many. Water was available at all times throughout the experiment. All subjects were

tested in a familiar testing room with a familiar experimenter.

2.2. Apparatus

A rectangular table was painted green and placed into a rectangular testing booth

with three transparent Plexiglas sides (Fig. 1). The table (84 32 45) had sliding

food dishes (12 8 2.5 cm) attached to either side which could travel 35 cm. In

Table 1

Sex, age, and rearing history of participating chimpanzee subjects

Name Sex Age Rearing history

Robert Male 26 Human

Reit Female 25 Human

Sandra Female 9 Mother

Jahaga Female 9 Mother

Frodo Male 9 Mother

Fifi Female 9 Mother

Truddy Female 9 Mother

Patrick Male 5 Mother



5/20

addition, a plastic tube (4 cm in diameter and 57 cm long) extended 31 cm from

underneath the table top into the wire mesh below the table and at the front of

the testing booth (98 95 cm). Fruit juice could be poured through the tube to

attract the subjects to sit in front of the center glass window (83 48 cm) across

the table from the experimenter. In the booth on either side of the table were Plex-iglas testing windows (75 50 cm) that each had small oval holes cut into them

(10 5 cm) which subjects could reach their fingers through to obtain food from

the food dishes when pushed within reach of the subject.

2.3. Procedure

2.3.1. Introduction

The subjects had never been tested with the apparatus, general testing procedure,

or competitive humans (i.e., human caretakers almost exclusively share food altruis-

tically). Therefore, to acquaint the subjects to the apparatus and procedure, they first

received four non-competitive introduction trials. E attracted the subject to sit acrossthe table from him, behind the center window, by pouring juice for the subject

through the juice tube which she could drink with her mouth. Once the subject

was sitting behind the center window drinking juice, pieces of banana were placed

on both food dishes and pushed within reach of the subject (Fig. 1). Subjects quickly

learned that they could obtain both pieces of banana by reaching through the oval

hand holes in the Plexiglas testing windows on either side of the table.

Second, E changed his behavior and began to compete with the subject. The pro-

cedure was the same as the initial non-competitive trials with the exception that E

Fig. 1. The experimental apparatus showing the experimental room, the testing booth, experimenter, and

subject (in addition, the placement of the food table, the sliding food trays, the hand holes in the two

Plexiglas windows for the subject to reach through, and the juice tube are also visible. Note. The tube only

protruded from the booth a few centimeters and could not be held by the subject. Therefore, subjects

always drank out of the juice tube using their mouth forcing them to sit facing the experimenter).



6/20

pulled the food out of the subjects reach if they attempted to approach the food (E

always stared straight ahead). In addition, if the subject refrained from approaching

for over 5 s then the food was removed. In this way the subjects would understandthat E, like their group mates, no longer was willing to share prized food (i.e.,

banana pieces) with them. To help assure that subjects understood that E was

now a competitor like their group mates, this competitive introduction procedure

was repeated until either: (1) the subject refrained from approaching for four out

of five consecutive trials or (2) they received 36 warm-up trials.

2.3.2. Test sessions

Once each subject had been introduced to the new apparatus, general procedure,

and competitive E, subjects were tested in three different tests each with its own

experimental setup. Each test consisted of three different conditions, all of which fol-

lowed the same general testing procedure used in the introduction. Two conditions in

each test were experimental while the third was a non-social control designed to mea-

sure subjects baseline behavior with the apparatus but without the experimenter.

Subjects were rewarded differently in each of the three types of conditions within

a test session. Once the subject was sitting across the testing table from E, drinking

juice with their mouth and E had pushed the food within reach of the hand holes, the

following conditions and rewarding were used for each of the three sessions.

2.3.2.1. Test 1: Body orientation (Fig. 2). Face and chest condition. E oriented his face

and chest towards one of the two pieces of food, with his back turned to the other. If

subjects approached, they were allowed to retrieve the piece of food behind the

experimenter, but the food was removed if the subjects approached the food Es face

and chest were oriented toward.Face versus chest condition. E oriented his face toward one of the food dishes while

orienting his chest toward the other piece of food. Regardless of which piece of food

the subject approached, E removed the food.

Non-social control. E immediately left the testing room allowing the subject to

obtain both pieces of food.

2.3.2.2. Test 2: Occluder (Fig. 3). Face and chest condition. Same as in Session 1.

Occluder condition. A plastic occluder (74 50 cm) was placed over one of the

choice windows between trials. The occluder also had a hole (10 5 cm) allowing

subjects to reach through the hand hole in Plexiglas window to obtain the food.

Regardless of which piece of food the subject approach, E removed the food whilestaring straight ahead.

Non-social control. The occluder was placed on one of the Plexiglas windows and

E left, thus allowing the subject to retrieve both pieces of food.

2.3.2.3. Test 3: Split occluder (Fig. 4). Double occluder condition. Two occluders were

placed on the same side of table. One of these occluders covered one of the Plexiglas

windows (74 50 cm) while the other barrier (71.5 50 cm) was placed upright on

the table. The subject was allowed to retrieve food from behind the occluders, while



7/20

food was moved out of reach if the subject approached the unoccluded Plexiglas win-

dow (E stared straight ahead).

Split occluder condition. A barrier was placed on either side of the table. Each

side had barriers that equalled half the occluding surface area of the barriers used

in Double Occluder condition. One barrier (72 50 cm) was placed on one side of

the table and acted as an occluder. Meanwhile, two barriers (each 72 25 cm),

that were each half the size of the occluder, were placed on the other side of

the table. One of the two smaller barriers was attached to the Plexiglas window

at eye level while the other was placed on the table. In addition, regardless of

which piece of food the subject approached, E removed the food and staredstraight ahead.

Non-social control. The barriers were positioned as in the Split Occluder condi-

tion. In this condition E left allowing the subject to obtain both pieces of food.

2.4. Design

Each test was completed on a separate day, in the order 13, and all followed

the same general design. Each test session contained two introduction trials (to

Fig. 2. A picture of experimental manipulations used in the Test 1 and the mean number of approaches

within each of the three test conditions (note: the two social cues in the two social tests were presented on

the right side (as in pictures) in half of trials and on the left side in the other half of trials).



8/20

remind subjects before starting that, unlike usual, E was a competitor) and 10 tri-als of each of the different conditions (30 total trials). The conditions were coun-

terbalanced across subjects using two different orders with half of the subjects

receiving each of the two orders in each test. Within each test session the condi-

tions were presented to each subject in a mixed order in 10 sets of three trials.

Each set of three trials contained one of each type of trial being tested in that

session (trials were counterbalanced to assure that subjects had the opportunity

to be rewarded regularly for hiding so that their motivation for competing against

E might be maintained across the entire session). Each condition was counterbal-

anced for the side on which each manipulation was administered (e.g., in the face

and chest condition E oriented both toward the food on the left and the right for

half of the trials, etc.).Finally, as a supplemental control, after all three of the test sessions were complet-

ed, all subjects were tested in an additional control condition in which barriers were

present but did not occlude Es view of the subjects approach to either food dish. As

in the Split Occluder condition, an opaque barrier was placed on either side of the

testing table. One barrier (35 25 cm) was half the size of the other (72 25 cm),

but neither barrier was large enough to act as a visual occluder. The procedure

was identical to the previous tests with the exception that subjects were rewarded

for approaching behind the larger of the two non-occluding barriers. Each subject

Fig. 3. A picture of experimental manipulations used in Test 2 and the mean number of approaches within

each of the three test conditions.



9/20

participated in a test session of 10 consecutive trials with the placement of the bar-

riers (left or right) being counterbalanced within and between subjects. This supple-

mental control is designed to test if subjects prefer larger barriers, regardless of their

occluding properties, while competing with E.

2.5. Scoring and analysis

All trials were coded from video for whether subjects approached (moved within

reaching distance of food) within the 30 s trial of one of the pieces of food (left or right).

All of the experimental trials werealso coded for whether the subject approached on the

side where it was most difficult for E to see her. This means a trial was scored as hidingwhen: Face and Chest the subject approached from behind Es back; Face versus

Chest the subject approached towards Es chest; Occluder the subject approached

behind the occluder; Double Occluder the subject approached behind the two occlud-

ers; Split Occluder the subject approached behind the solid occluder. In addition, all

trials were coded for whether the approach was direct (Fig. 5: subjects first reduced the

distance between themselves and the food) or indirect (Fig. 5: subjects first distanced

themselves from the food 13 m before reducing the distance between themselves

and the food). Twenty percent of trials were coded by a second coder who was blind

Fig. 4. A picture of experimental manipulations used in Test 3 and the mean number of approaches within

each of the three test conditions.



10/20

to the condition. Interobserver reliability was 99.7% for approaches, with a Cohens j

of .98 and 97.9% for the type of approach with a Cohens j of .86.

To test the predictions of the competing hypotheses, we planned to analyze sub-jects approach behavior in several ways. Within each condition a paired t-test was

used to compare the number of trials subjects, as a group, hid their approach or

did not (percentage of approaches are presented in all figures because of differences

in the number of trials in which subjects chose to approach at all during each ses-

sion). In addition, to assess individual hiding preferences each subjects combined

hiding preference in all six social conditions was compared to chance (each subject

generally approached too infrequently in each session to provide sufficient power

for reliable assessment of individual preferences in each of the six social conditions

separately). Finally, we examined whether as a group, subjects approached indirectly

more often in the experimental than control condition in each test using a one-way

ANOVA and paired comparisons (Fischer LSD test corrected for multiple planned

comparisons). Because this experiment was designed to test the directional predic-

tions of the visual perspective-taking hypothesis against the null predictions of the

contextual and learned cues hypothesis, all statistics were one tailed unless otherwise

noted.

In order to assess if subjects came to the experiment with an understanding of how

to hide from Es view, or if they learned to do so during the experiment, we conduct-

ed three tests of learning. First, within each test session we looked to see if subjects

approach behavior changed across the session. The groups approach preferences in

the social conditions of each session were compared using (1) trials 13 compared to

trials 46, and (2) the first 10 trials to the last 10 trials using paired t-tests (multiple

Fig. 5. An illustration of a subjects (1) direct approach and (2) indirect approach to the left of the

experimenter. When approaching directly subjects: (1a) sat at the juice tube out of the camera s view,

(1b) look directly around the corner of the test booth, (1c) approached directly around the corner of the

test booth, (1d) and reached for the food. When approaching indirectly subjects: (2a) sat at the juice tube

out of view of the camera, (2b) distanced themselves from the food and experimenter, (2c) returned out of

view of the experimenter (notice occluded window), (2d) and reached for the food.



11/20

tests were not corrected since such tests favor finding evidence of learning). Second,

we tested to see if subjects approached and hid more in the conditions where they

were rewarded than those in which they were not using a paired t-test. Third, we alsoexamined subjects first approach behavior. This was done by comparing the groups

preferred direction of approach in the first trial of the two experimental conditions in

each test. For the first two learning analysis we had no directional prediction since

during a test session subjects might have increased hiding due to rewarding or

decreased hiding due to a loss of motivation. The third analysis was one tailed since

the visual perspective-taking hypothesis makes the directional prediction that sub-

jects will hide within trials.

3. Results

In the introduction session seven of the eight subjects met the criteria and quickly

refrained from approaching (mean 13 trials; range 225 trials). One subject (Truddy)

continued approaching throughout the 36-trial session, but because the introduction

was only designed to introduce the subjects to the competitive E (not act as a crite-

rion for participation or as a behavior shaping procedure) this subject also partici-

pated in all subsequent tests (she concealed her approach in a majority of trials in

five of the six experimental tests).

In Test 1: Body Orientation, as a group, subjects avoided approaching the food

that E was watching in both the Face and Chest condition, [t (7) = 3.87,

p = 0.003], and in the Face versus Chest condition, [t (7) = 2.65, p = 0.017]. As a

group, subjects did not have a significant preference for approaching one side of

the booth or the other in the Non-social control condition (Fig. 2).In Test 2: Occluder, as a group, subjects again had a preference for avoiding the

food E was watching in the Face and Chest condition [t (7) = 3.38, p = 0.006], rep-

licating the results from the first test. In the Occluder condition, as a group, subjects

did not have an overall preference for approaching behind either the opaque occlud-

er or the transparent barrier [t (7) = 1.31, p = NS] (Fig. 3). However, this result is

explained by the fact that some subjects had a strong aversion to the occluder regard-

less of whether E was present or not as revealed in the Non-social control in which

overall subjects preferred to approach behind the transparent barrier [t (7) = 2.86,

p = 0.012, two tailed].

Because of this unexpected preference for the transparent barrier in the Non-

social control, we ran additional analysis of this test to examine more closely if sub-jects used the occluder to hide. First, a two-way ANOVA (condition barrier type)

was used to compare subjects approach behavior in the Occluder condition to that

in the Non-social control. There was a significant effect of condition [F(1,7) = 16.66,

p = 0.005] and a significant condition barrier type interaction [F(2,21) = 15.234,

p = 0.006], with subjects avoiding the transparent barrier significantly more in the

Occluder condition than in the Non-social control. Second, we also examined the

first trials within the Occluder test and Non-social control conditions. Subjects

had a significant preference for approaching behind the occluder in the first trial they



12/20

approached in the Occluder condition but not in the Non-social control [Sign Test,

p < 0.05]. Finally, after removing the two subjects who had significant individual

preferences for the transparent barrier in the Non-social control (both individualsapproached behind the transparent barrier in all 10 trials, p < 0.002, binomial prob-

ability, two tailed), we found that the remaining six subjects did have an overall pref-

erence for hiding behind the occluder in the Occluder condition (T= 19, N= 6,

p < 0.045, Wilcoxon Sign Rank test).

In Test 3: Split Occluder subjects preferred to approach behind the visual occluder

both in the Double Occluder condition [t (7) = 3.56, p = 0.005], and in the Split

Occluder condition [t (7) = 3.59, p = 0.005]. Subjects had no preference for either

of the different types of barriers within the Non-social control (Fig. 4).

Table 2 presents the percentage of trials in which each subject hid in the six dif-

ferent social conditions (percentages were used because subjects did not always

approach in every trial). With one exception (Robert), all subjects hid their approach

in over two-thirds of trials in at least five of the six social conditions (this was the

case in all conditions for three subjects). In addition, half of the subjects hid their

approach in a majority of trials (i.e., over half) in all six social conditions.

Table 2 also presents the percentage of trials in which each subject hid when all six

social conditions are combined. All but one subject (Robert) hid from their compet-

itor at above chance levels (curiously, Robert, the only adult male tested, initially hid

in each test session, but then seemed to switch strategies and instead displayed in

sight of the male experimenter or begged for the food using a food begging gesture

before attempting to retrieve the visible food much in the way that male chimpanzees

do in the wild when trying to obtain prized food such as meat from one another).

Table 2

Percentage of trials in which subjects hid from the experimenter in each of the six social conditions and all

six social conditions combined

Subject Condition Total

Test 1: body orientation Test 2: occluder Test 3: split occluder

Face and

chest

Face versus

chest

Face and

chest

Occluder Double occluder Split occluder

Robert 40 50 100 25 0 100 52.5

Reit 100 75 50 100 100 100 87.5*

Sandra 71 67 67 100 100 70 79.1***

Jahaga 90 75 80 80 100 90 85.6***

Frodo 100 83 100 70 100 60 85.5***

Fifi 80 33 80 75 80 67 69.2**

Truddy 80 75 90 14 67 75 66.7*

Patrick 90 75 67 100 90 56 79.5***

Total 81.4 66.7 79.2 70.5 79.6 77

1st trial 6 6 5 7 6 7

Symbols indicate significant overall hiding preference (binomial probability: *p < 0.05, **p < 0.01,***p < 001). Bold type represents conditions in which subjects hid on their first trial in which they

approached. Also shown is the number of subjects that hid in the first trials across conditions.



13/20

With respect to the use of indirect approaches, analyses revealed that in the exper-

imental conditions of Test 1, six of eight subjects approached indirectly in one or

more trials (a mean of 15% of experimental trials), whereas no subject approachedindirectly in the Non-social control. The conditions thus differed from one another

in the number of indirect approaches [F(2,21) = 3.63, p = 0.044], with both of the

experimental conditions in Test 1 having more indirect approaches than the Non-

social control (p < 0.05 in both cases, Fischer LSD). In Test 2, five of eight subjects

approached indirectly in one or more trials of the Experimental conditions (a mean

of 16% of experimental trials), whereas no subject approached indirectly in the Non-

social control. This difference in indirect approaches neared significance

[F(2,21) = 2.65, p = 0.094], with the Occluder condition having significantly more

indirect approaches than the Non-social control (p < 0.05, Fischer LSD). Subjects

never used an indirect approach in Test 3 in which it would have served no purpose

(likewise, in the introduction only one subject approached indirectly only one time,


14/20

group, did not have a preference for approaching the larger barrier (rewarded) or the

smaller barrier (unrewarded) on their first trial or over the entire session [t (7) = 1.51,

p = NS] with only three of eight subjects approaching behind the larger barrier(rewarded) in more trials than the smaller barrier (unrewarded).

4. Discussion

The results of these experiments demonstrate that chimpanzees can flexibly use

knowledge of what a competitor can and cannot see to develop active, deceptive

strategies for concealing their approach to contested food and they do this from

the very first trials in several novel situations. First, contrary to the conclusions of

Povinelli and colleagues (Povinelli et al., 1999; Povinelli & Eddy, 1996; Reaux

et al., 1999; Theall & Povinelli, 1999) based on studies using a different methodology,

chimpanzees in the current experiment spontaneously avoided food that the experi-

menter was watching, as indicated by gaze direction, and instead approached food

that he was not watching, even when the majority of his body (including his arms

and hands which can physically remove food from the subjects reach) was oriented

in this same direction. Second, contrary to the conclusions of Karin-Karin-DArcy

and Povinelli (2002) and consistent with those of Hare et al. (2000, 2001), subjects

in the current experiment preferred to approach food behind various visual occluders

while refraining from approaching food behind non-occluding transparent or opa-

que barriers. In addition, in the supplemental control session when subjects were

forced to choose between two non-occluding barriers, they did not prefer to

approach food near the larger of the two non-occluding barriers even though they

were rewarded for doing so.However, perhaps the most striking aspect of subjects behavior was their exclu-

sive use of indirect approaches when initially distancing themselves from the food

could potentially aid in concealing their later approach towards the food. Subjects

only used indirect approaches if the experimenter was able to see them distance

themselves from the food, but subsequently could not see them approach the food

after positioning themselves behind the experimenter or an occluder (Test 1 and

2). Meanwhile, subjects did not use indirect approaches if the positioning of the

occluders prevented the experimenter from seeing them distance themselves from

the food (Test 3), or they could easily be seen approaching after distancing them-

selves from the food (introduction). This indirect approach behavior is striking

because it suggests the possibility that subjects not only understood it was importantto be hidden from their competitors view while approaching contested food, but that

they also understood that in some cases it was useful to hide their attempt to hide.

These results are thus consistent with the predictions of the visual perspective-taking

hypothesis, while providing little, if any, support for the predictions of the learned

cues hypothesis (i.e., subjects have learned to react to a set of invariant perceptual

regularities).

Meanwhile, while the subjects approach behavior is consistent with the

predictions of the visual perspective-taking hypothesis, the design of the current



15/20

experiment makes it difficult to account for such results by proposing a task specific

behavior reading or learning hypotheses (e.g., Heyes, 1993, 1998). While the Face

versus Chest condition (Test 1) demonstrates that chimpanzees can take advantageof a relatively subtle behavioral cue such as gaze direction in assessing whether a

competitor will see them, subjects did not rely solely on such behavioral cues when

hiding. Throughout the second and third tests the Occluder, Double Occluder, and

Split Occluder conditions the experimenter always faced directly forward (remain-

ing motionless until subjects made a choice), yet subjects still spontaneously used the

barriers to conceal their approach. Moreover, it is also difficult to argue that sub-

jects preferences represent a set of simple contingencies learned within the experi-

ment. First, subjects showed their hiding preferences in the first trial of both

experimental conditions within each of the three tests. Second, subjects maintained

their initial hiding preferences throughout the session in the Face versus Chest con-

dition, Occluder condition (see additional analysis), and Split Occluder condition,

even though subjects were never successful in obtaining food in these unrewarded

conditions. Third, there was no evidence that subjects rapidly increased their hiding

response in the first six trials of each test or that they more gradually increased their

hiding by the second half of each test. Finally, although subjects approached and hid

more in the rewarded conditions than in the non-rewarded conditions, there is no

evidence that subjects level of hiding in the rewarded and unrewarded conditions

changed relative to each other across the first and second half of the experiment

(i.e., this is contrary to a within-session learning hypothesis that predicts that in

the second half of the experiment subjects hiding will decrease in the unrewarded

conditions while increasing in the rewarded condition). Further the fact that the sub-

jects hid slightly more in the rewarded conditions is easily explainable by the fact that

the choices in those conditions were intentionally designed to be less ambiguous thanthe unrewarded conditions (i.e., to maintain motivation). As an example, the biggest

difference between a rewarded and unrewarded condition in hiding is in Test 1 where

subjects hid 15% more in the Face and Chest condition than in the Face versus Chest

condition. However, in the Face and Chest condition the subjects likely hid more

simply because the experimenter in this condition both could not see one of the food

pieces and had to turn around to remove this same piece of food (see Fig. 2). Mean-

while in the Face versus Chest condition the subject had to choose between hiding

from view or avoiding the food that was easier for the experimenter to remove

making the choice more ambiguous regardless of rewarding. Therefore, overall there

is little reason to believe that the rewarding scheme caused the subjects spontaneous

hiding preferences to change significantly during the relatively short experimentalsessions. Instead, it seems most plausible to conclude that subjects flexibly general-

ized previous cognitive skills in solving all of the novel social problems with which

they were presented in the current experiment arguably representing a successful

case of triangulation (Heyes, 1993, 1998).

However, there is one specific egocentric hypothesis that must be addressed in

relation to the current results. It might be suggested that throughout ontogeny

and into adulthood chimpanzees simply learn that contested food is best obtained

when a path is taken to food that prevents them from seeing their competitors face



16/20

as they approach. This would mean that subjects in the current experiment hid

behind visual occluders not because they understood the experimenter could not

see them, but instead because they knew if they went behind the occluder they couldnot see the experimenter. At first this might seem to be a useful candidate explana-

tion for the current results, but details of the experimental design, the subjects

behavior, and previous findings make such an explanation seem implausible.

First, the Split Occluder condition directly addresses this alternative hypothesis.

Recall in this condition that one piece of food was behind a solid barrier while the

other was behind a split barrier (one on the testing table and the other at eye level:

see Fig. 4). Therefore, in both cases subjects could potentially hide the experimenters

face from their own view as they retrieved the food, and so in this condition the

avoid face hypothesis predicts no preference in subjects approaches. Yet subjects

preferred to steal the food from behind the solid barrier, not the split barrier, con-

sistent with the idea that they understood the human could see them behind the split

barrier even if they could position themselves so as not to see the humans face. Sec-

ond, in the Face versus Chest condition, the human experimenters face (including

eyes) and body were always visible regardless of which testing window subjects

approached (see Fig. 2). Therefore, the experimenter was completely visible to the

subjects regardless of which direction subjects approached, yet they preferred to

avoid the food at which the experimenter was gazing (and they could not only rely

on the rule, do not approach when the eyes of another are directed towards food

since gaze direction could not help them hide themselves in Tests 2 and 3 when the

experimenter always stared directly at the subjects starting position). Third, the

avoid face hypothesis predicts only that subjects will take a direct route to hide

the competitor from their own view. Yet, the subjects often took an indirect, circu-

itous route to the food and only when their departure away from the food could beseen and their return would remain hidden. Finally, the avoid face hypothesis can-

not account for previous results in which chimpanzees show an understanding of the

occluding properties of opaque barriers even when their competitor is out of sight

(Hare et al., 2000, 2001). Thus, it seems plausible that in the current experiment sub-

jects are using a similar understanding in deciding how best to hide themselves from

their competitors view. However, in the future experiments should be designed to

test the prediction of the avoid face hypothesis further.

The current results, therefore, appear to replicate and extend other recent findings

that chimpanzees understand others visual perception in the form of level 1 visual

perspective-taking (Flavell, 1992). Studies of chimpanzee gaze following suggest that

they may know when others can see things that they themselves cannot; for example,when chimpanzees gaze follow but do not see anything of interest, they look back to

the gazer and follow her gaze a second time (Call et al., 2000) this checking back

being a common criteria in assessing human childrens understanding of visual per-

ception. If a chimpanzee is subsequently unable to detect anything of interest, they

quickly habituate to the gaze of their social partner (Tomasello et al., 2001). If a visu-

al barrier prevents a chimpanzee from matching their line of sight with that of anoth-

er individual, they will move in order to see what the other individual can see

(Povinelli & Eddy, 1996; Tomasello et al., 1999). In addition, chimpanzees ignore



17/20

interesting objects if their social partner is not gazing in the direction of the distrac-

tion (Tomasello et al., 1999). Moreover, the Hare et al. (2000, 2001) food competi-

tion studies suggest that chimpanzees know what others can and cannot see and evenwhat others have seen or have not seen in the past. Meanwhile, the current results

add a more active dimension to these findings, showing that chimpanzees can predict

and manipulate what others do and do not see.

However, actively manipulating what another will be able to see does more than

support the visual perspective hypothesis. Subjects hiding behavior also provides

some of the strongest evidence that chimpanzees are capable of intentional decep-

tion. Although a number of authors (Goodall, 1986; Hirata & Matsuzawa, 2001;

Menzel, 1974) have reported that chimpanzees often distance themselves from hid-

den food in the presence of dominants, prior findings could be explained if subjects

simply learned over time to adjust their approach behavior in response to the

behavior of their competitor (i.e., if the subordinate moves towards the hidden

food so does the dominant). However, in the current study the hiding and indirect

approaches cannot be characterized as adjustments to the competitors behavior

given that the competitive experimenter remained motionless while a subject made

its choice. In addition, further support for interpreting the approach behavior of

subjects as intentional deception is the frequency of indirect approaches. A number

of authors have pointed out that there are constraints on the frequency for which

deception can occur. To avoid detection and costly punishment it has been argued

that rates of intentional deception will remain low in social species such as prima-

tes (Cheney & Seyfarth, 1990; Hauser, 1992). In this regard, it is interesting to note

that subjects from the current experiment only attempted to approach indirectly,

the most deceptive of approach behavior, in a minority of trials. Therefore, it is

possible that the subjects use of indirect approaches represents an experimentaldemonstration of intentional deception in chimpanzees subjects were trying to

hide their attempt to hide and did so infrequently to avoid detection of their decep-

tive strategy.

If it is the case that subjects hiding and use of indirect approaches represents a

case of intentional deception, it is even more difficult to know whether to character-

ize such behavior as a case of withholding or falsifying information (Cheney &

Seyfarth, 1990). Subjects may have simply been trying to withhold visual informa-

tion about their approach or, as might be suggested by their indirect approaches,

they may even have been trying to provide false information about the ultimate goal

of their movements. It is possible that the chimpazees deceptive behavior is best

characterized as some intermediate level of deception such as active concealment(i.e., subjects were trying to block their competitors informational access). Unfortu-

nately, the current results do not give us the resolution needed to determine more

precisely on which deceptive level chimpanzees are capable of operating. Instead,

the current results do more to suggest the need for future research on chimpanzee

deception to clarify such issues. For example, it remains unclear exactly how flexible

chimpanzees are in creating and deploying various deceptive strategies. Perhaps,

chimpanzees are only deceptive in a very narrow set of context (i.e., hiding visually

during food competition) suggesting they have little understanding of how to effective-



18/20

ly manipulate their environment or behavior in order to affect others psychological

states. And of course there are still types of deception for example, active falsifica-

tion and deception involving language (i.e., lying) that may prove to be the soleprovince of human beings (Call & Tomasello, 1999).

Therefore, given the need for more experimental work on chimpanzee decep-

tion, one promising direction for future investigations will be research aimed at

testing whether chimpanzees can also assess and manipulate the perceptual states

of others across modalities. Given observational evidence from the wild (Hauser,

1990; Watts & Mitani, 2001) and experimental evidence from captivity (Hauser

et al., 1993) and the wild (Wilson, Hauser, & Wrangham, 2001) that chimpanzees

withhold vocalizations (or avoid making loud noise in general) during contested

matings, food competition, and territorial disputes, one future direction might

be to use the current experimental paradigm to assess whether chimpanzees also

know when others can and cannot hear them approaching contested food. In

addition, another possibility would be to develop a new experimental paradigm

to investigate whether chimpanzees also actively falsify vocalizations in order to

intentionally deceive others.

Developing such non-verbal tests of social cognition in our closest primate rela-

tive will play a central role in not only testing a number of theories relevant to the

evolution of human social cognition (e.g., Krebs & Dawkins, 1984; Povinelli,

2000; Tomasello & Call, 1997; Whiten, 1994), but will also directly test the predic-

tions of a number of theories for human social cognitive development (e.g., Asting-

ton & Jenkins, 1999; Smith, 1996; Tomasello, 1999). Therefore, as experimental

methods continue to be improved work with non-humans will continue to become

increasingly relevant to a number of fields within the Cognitive Sciences (Hauser

& Fitch, 2002).

Acknowledgments

We thank the Wolfgang Koehler Primate Research Center animal caretakers, Da-

vid Buttelmann, Esther Herrmann, and Alicia Perez Melis for tremendous help with

data collection and reliability coding. We thank Jochen Barth for his clever proce-

dural suggestions. We also thank Marc Hauser, Matthew McIntyre, Cory Miller,

Daniel Stahl, and several anonymous reviewers for very constructive comments on

an earlier version of the manuscript. Finally, we thank Cara Bean for the nice illus-

tration used in Fig. 1.

References

Anderson, J. R., Kuroshima, H., Kuwahata, H., Fujita, K., & Vick, S.-J. (2001). Training squirrel

monkeys (Saimiri sciureus) to deceive: Acquisition and analysis of behavior toward cooperative and

competitive trainers. Journal of Comparative Psychology, 115(3), 282293.

Astington, J. W., & Jenkins, J. M. (1999). A longitudinal study of the relation between language and

theory-of-mind development. Developmental Psychology, 35, 13111320.

http://-/?-http://-/?-


19/20

Call, J., Agnetta, B., & Tomasello, M. (2000). Cues the chimpanzees do and do not use to find hidden

objects. Animal Cognition, 3(1), 2334.

Call, J., & Tomasello, M. (1999). A nonverbal theory of mind test. The performance of children and apes.

Child Development, 70, 381395.

Cheney, D. L., & Seyfarth, R. M. (1990). How monkeys see the world: Inside the mind of another species.

Illinois: University of Chicago Press, 377pp.

Coussi-Korbel, S. (1994). Learning to outwit a competitor in Mangabeys. Journal of Comparative

Psychology, 108(2), 164171.

de Waal, F. (1998). Chimpanzee politics: Power and sex among apes (Vol. xv, rev. ed., 235pp).

Flavell, J. H. (1992). Perspectives on perspective taking. In H. Beilin & P. B. Pufall (Eds.), Piagets theory:

Prospects and possibilities (pp. 107139). Hillsdale, NJ: Lawrence Erlbaum Associates.

Fujita, K., Kuroshima, H., & Masuda, T. (2002). Do tufted capuchin monkeys (Cebus apella)

spontaneously deceive an opponent? A preliminary analysis of an experimental food-competition

contest between monkeys. Animal Cognition, 5, 1925.

Goodall, J. (1986). The chimpanzees of gombe. Cambridge: Belknap Cambridge.

Gygax, L. (2000). Hiding behaviour of long-tailed macaques (Macaca fascicularis): II. Use of hiding places

during aggressive interactions. Ethology, 106(5), 441451.Hare, B. (2001). Can competitive paradigms increase the validity of social cognitive experiments on

primates? Animal Cognition, 4, 269280.

Hare, B., Call, J., Agnetta, B., & Tomasello, M. (2000). Chimpanzees know what conspecifics do and do

not see. Animal Behaviour, 59(4), 771785.

Hare, B., Call, J., & Tomasello, M. (2001). Do chimpanzees know what conspecifics know? Animal

Behaviour, 61(1), 139151.

Hare, B., & Tomasello, M. (2004). Chimpanzees are more skillful in competitive than in cooperative

cognitive tasks. Animal Behaviour, 68, 571581.

Hauser, M. D. (1990). Do chimpanzee copulatory calls incite malemale competition? Animal Behaviour,

39(3), 596597.

Hauser, M. D. (1992). Costs of deception: Cheaters are punished in rhesus monkeys (Macaca mulatta).

Proceedings of the National Academy of Sciences of the United States of America, 89 , 1213712139.

Hauser, M. D. (1997). Minding the behaviour of deception. In A. Whiten & R. W. Byrne (Eds.),

Machiavellian intelligence II: Extensions and evaluations (Vol. xii, pp. 112143), 403pp.Hauser, M., & Fitch, T. (2002). What are the uniquely human components of the language faculty? In M.

Christiansen & S. Kirby (Eds.), Language evolution. Oxford, England: Oxford University Press.

Hauser, M. D., Teixidor, P., Field, L., & Flaherty, R. (1993). Food-elicited cells in chimpanzees: Effects of

food quantity and divisibility. Animal Behaviour, 45(4), 817819.

Heyes, C. M. (1993). Anecdotes, training, trapping and triangulating: Do animals attribute mental states?

Animal Behaviour, 46(1), 177188.

Heyes, C. M. (1998). Theory of mind in nonhuman primates. Behavioral and Brain Sciences, 21(1),

101134.

Hirata, S., & Matsuzawa, T. (2001). Tactics to obtain a hidden food item in chimpanzee pairs ( Pan

troglodytes). Animal Cognition, 4(34), 285295.

Hostetter, A., Contera, M., & Hopkins, W. (2001). Differential use of vocal and gestural communication

by chimpanzees (Pan troglodytes) in response to the attentional status of a human (Homo sapiens).

Journal of Comparative Psychology, 115(4), 337343.

Kaminski, J., Call, J., & Tomasello, M. (2004). Body orientation and face orientation: Two factors

controlling apes begging behavior from humans. Animal Cognition, 7(4), 216233.

Karin-Karin-DArcy, M. R., & Povinelli, D. J. (2002). Do chimpanzees know what each other see. A

closer look. International Journal of Comparative Psychology, 15(1), 2154.

Krebs, J. R., & Dawkins, R. (1984). Animal signals: Mind reading and manipulation. In J. R. Krebs & N.

B. Davies (Eds.), Behavioural ecology: An evolutionary approach (pp. 380401). Oxford, England:

Blackwell.

Kummer, H., Anzenberger, G., & Hemelrijk, C. (1996). Hiding and perspecttive taking in long-tailed

macques (Macaca fascicularis). Journal of Comparative Psychology, 110, 97102.



20/20

Leavens, D., & Hopkins, W. (1998). Intentional communication by chimpanzees a cross-sectional study

of the use of referential gestures. Developmental Psychology, 34(5), 813822.

Menzel, E. (1974). A group of young chimpanzees in a one-acre field: Leadership and communication. In

A. M. Schrier & F. Stollnitze (Eds.), Behavior of nonhuman primates (pp. 83153). New York:

Academic Press.

Mitchell, R. W., & Anderson, J. R. (1997). Pointing, withholding information, and deception in capuchin

monkeys (Cebus apella). Journal of Comparative Psychology, 111(4), 351361.

Myowa-Yamakoshi, M., Tomonaga, M., Tanaka, M., & Matsuzawa, T. (2003). Preference for human

direct gaze in infant chimpanzees (Pan troglodytes). Cognition, 89(2), B53B64.

Povinelli, D. J. (2000). Folk physics for apes: The chimpanzees theory of how the world works . Oxford:

Oxford University Press.

Povinelli, D. J., Bierschwale, D. T., & Cech, C. G. (1999). Comprehension of seeing as a referential act in

young children, but not juvenile chimpanzees. British Journal of Developmental Psychology, 17(Pt. 1),

3760.

Povinelli, D. J., & Eddy, T. J. (1996). What young chimpanzees know about seeing. Monographs of the

Society for Research in Child Development, 61(3), vvi, 1152.

Povinelli, D. J., & Giambrone, S. (2001). Reasoning about beliefs: A human specialization? Commentaryon Meta-analysis of theory-of-mind development: The truth about false belief. Child Development,

72(3), 691695.

Povinelli, D. J., & Vonk, J. (2003). Chimpanzee minds: Suspiciously human?. Trends in Cognitive Sciences

7(4), 157160.

Reaux, J. E., Theall, L. A., & Povinelli, D. J. (1999). A longitudinal investigation of chimpanzees

understanding of visual perception. Child Development, 70(2), 275290.

Ristau, C. A. (Ed.). (1991). Cognitive ethology: The minds of other animals: Essays in honor of Donald R.

Griffin (Vol. xx, 332pp).

Shettleworth, S. J. (1998). Cognition, evolution, and behavior. Oxford: Oxford University Press.

Smith, P. (1996). Language and the evolution of mind-reading. In P. Carruthers & P. Smith (Eds.),

Theories of theories of mind (pp. 344354). Cambridge, England: Cambridge University Press.

Theall, L. A., & Povinelli, D. J. (1999). Do chimpanzees tailor their gestural signals to fit the attentional

states of others?. Animal Cognition 2(4), 207214.

Tomasello, M. (1999). The cultural origins of human cognition (Vol. vi, 248pp.).Tomasello, M., & Call, J. (1997). Primate cognition. New York, NY, USA: Oxford University Press, 517.

Tomasello, M., Call, J., & Hare, B. (2003a). Chimpanzees understand psychological states The question

is which ones and to what extent. Trends in Cognitive Sciences, 7(4), 153156.

Tomasello, M., Call, J., & Hare, B. (2003b). Chimpanzees versus humans: Its not that simple. Trends in

Cognitive Sciences, 7(6), 239240.

Tomasello, M., Hare, B., & Agnetta, B. (1999). Chimpanzees, Pan troglodytes, follow gaze direction

geometrically. Animal Behaviour, 58(4), 769777.

Tomasello, M., Hare, B., & Fogleman, T. (2001). The ontogeny of gaze following in chimpanzees, Pan

troglodytes, and rhesus macaques, Macaca mulatta. Animal Behaviour, 61(2), 335343.

Watts, D., & Mitani, J. (2001). Boundry patrols and intergroup encounters in wild chimpanzees.

Behaviour, 138, 299327.

Whiten, A. (1994). Grades of mindreading. In C. Lewis & P. Mitchell (Eds.), Childrens early understanding

of mind: Origins and development. Hove, UK: Erlbaum.

Whiten, A., & Byrne, R. W. (1988). Tactical deception in primates. Behavioral and Brain Sciences, 11(2),

233273.

Wilson, M., Hauser, M., & Wrangham, R. (2001). Does participation in intergroup conflict depend on

numerical assessment, range location, or rank for wild chimpanzees? Animal Behaviour, 61, 12031216.

Woodruff, G., & Premack, D. (1979). Intentional communication in the chimpanzee: The development of

deception. Cognition, 7, 333362.


Documents

Chimpanzees Deceive a Human Competitor_06